--- license: cc-by-4.0 task_categories: - time-series-forecasting - feature-extraction - other tags: - radar - driver-monitoring - physiological-sensing - vision - hdf5 - in-cabin - multimodal - driver-vigilance --- # ICR-DS: In-Cabin Radar Driver State Dataset ICR-DS (In-Cabin Radar Driver State) is a multimodal naturalistic driving dataset for driver vigilance and physiological state monitoring. The dataset was collected from 28 participants during approximately 50 minutes of real-world urban and highway driving using synchronised wearable physiological sensing, radar, and an in-cabin RGB camera. To provide an objective measure of vigilance, each participant completed a Psychomotor Vigilance Test (PVT) before and after each drive. The dataset therefore supports longitudinal analysis of behavioural and physiological driver-state measurements, as well as future development of non-contact physiological monitoring using in-cabin radar sensing. --- ## Overview Each recording session is stored as a single HDF5 file. The file contains: - raw signals retained at native resolution - processed metrics computed from windowed aggregation - session-level attributes describing the participant, timestamps, sensor configuration, and modality availability - PVT summaries collected before and after the drive The dataset includes synchronised recordings from: - a 60 GHz FMCW radar sensor - a BioHarness 3 chest strap - an in-cabin RGB camera - session-level PVT outcomes OpenFace was applied to the video stream to derive eye and head-behaviour measurements. --- ## Session Attributes Each HDF5 file contains the following session attributes: | Attribute | Example | Description | |---|---|---| | session_id | "s027_2026-05-08" | Session identifier | | subject_id | "s027" | Subject identifier | | start_time_iso | "2026-05-08T11:27:21Z" | Session start time in UTC ISO-8601 format | | end_time_iso | "2026-05-08T11:28:32Z" | Session end time in UTC ISO-8601 format | | frame_period_s | 0.09 | Radar frame period in seconds | | carrier_freq_hz | 60000000000 | Radar carrier frequency | | wavelength_m | 0.005 | Radar wavelength | | range_resolution_m | 0.0843 | Radar range resolution | | has_radar | "present" | Radar availability flag | | has_vision | "present" | Vision availability flag | | has_bioharness | "present" | BioHarness availability flag | | has_gps | "present" | GPS availability flag | | has_pvt | "present" | PVT availability flag | --- ## HDF5 File Structure Each session file follows this structure: ``` / ├── raw/ │ ├── radar/ │ │ └── beamformed_cube │ ├── physiology/ │ │ ├── hr_1hz │ │ ├── br_1hz │ │ ├── ibi │ │ └── ibi_ts │ ├── vision/ │ │ ├── landmarks_raw │ │ ├── gaze_xy │ │ ├── pose_rxyz │ │ ├── eye_open │ │ └── confidence │ └── labels/ │ ├── motion_flag │ ├── predrive │ └── postdrive └── processed/ ├── physiology/ │ ├── mean_hr │ ├── mean_br │ ├── rmssd │ ├── sdnn │ ├── pnn50 │ └── hf └── vision/ ├── blink_rate ├── perclos ├── blink_duration ├── gaze_speed ├── gaze_x_std ├── gaze_y_std ├── head_speed ├── head_yaw_std ├── head_pitch_std └── head_roll_std ``` --- ## Radar Data Each radar sample is stored as a beamformed complex IQ tensor with shape: ``` (Window, IQ, Doppler, Direction, Range) = (200, 2, 5, 63, 7) ``` ### Radar tensor dimensions | Dimension | Size | Description | |---|---|---| | Window | 200 | Consecutive radar frames sampled every 90 ms, giving approximately 18 seconds of observation | | IQ | 2 | Real and imaginary components of the beamformed signal | | Doppler | 5 | Low-velocity Doppler bins centred on zero Doppler | | Direction | 63 | Beamformed look directions (9 azimuth × 7 elevation) | | Range | 7 | Selected range bins corresponding to the driver region | The raw radar stream is retained at native resolution and stored in `raw/radar/beamformed_cube`. --- ## Physiological Signals The BioHarness 3 chest strap provided: - heart rate at 1 Hz - breathing rate at 1 Hz - R-R intervals as event-based measurements Raw physiology is stored in: - raw/physiology/hr_1hz - raw/physiology/br_1hz - raw/physiology/ibi - raw/physiology/ibi_ts --- ## Vision Signals The in-cabin RGB camera was mounted next to the radar and focused on the driver's head and face. OpenFace was applied to the video stream to derive eye and head-behaviour measurements. Raw vision features are stored in: - raw/vision/landmarks_raw - raw/vision/gaze_xy - raw/vision/pose_rxyz - raw/vision/eye_open - raw/vision/confidence --- ## PVT Labels To provide objective vigilance measurements, participants completed a Psychomotor Vigilance Test before and after each drive. The following outcomes were extracted: - mean_rt_ms - median_rt_ms - reciprocal_rt_1_per_s - lapse_rate - commission_rate These are stored in: - raw/labels/predrive - raw/labels/postdrive --- ## Motion Flag A motion flag is stored in: - raw/labels/motion_flag This signal is generated from aligned vehicle and body motion sources and stored as a session-aligned derived label signal. --- ## Temporal Alignment Each modality had its own time stream and was synchronised using the latest common start time, with the earliest end time defining the shared analysis interval. Raw signals were retained at their native resolution. Feature extraction begins after a 5-minute offset to enable reliable HRV estimation and maintain alignment across modalities. --- ## Processed Features Processed features were computed from the raw signals using fixed temporal aggregation windows. ### Physiology Metrics Physiological features were computed every minute using a 5-minute rolling window. | Metric | Unit | Description | |---|---|---| | Heart rate | bpm | Mean heart rate | | Breathing rate | breaths/min | Mean breathing rate | | RMSSD | ms | Root mean square of successive R–R interval differences | | SDNN | ms | Standard deviation of normal-to-normal intervals | | HF power | ms² | High-frequency HRV power | | pNN50 | % | Percentage of successive R–R intervals differing by more than 50 ms | Stored in: - `processed/physiology/*` --- ### Vision Metrics Blink-based measures were computed over non-overlapping 1-minute windows. Gaze and head-pose features were computed over non-overlapping 10-second windows to capture short-term changes in visual scanning behaviour. #### Blink Metrics | Metric | Unit | Description | |---|---|---| | Blink rate | blinks/min | Number of blink events per minute | | Blink duration | s | Mean duration of detected blinks | | PERCLOS | % | Percentage of time the eyes are closed | Stored in: - `processed/vision/*` #### Gaze Metrics | Metric | Unit | Description | |---|---|---| | Gaze x std | rad | Variability of horizontal gaze angle | | Gaze y std | rad | Variability of vertical gaze angle | | Gaze speed | rad/s | Mean rate of gaze-angle change | #### Head Pose Metrics | Metric | Unit | Description | |---|---|---| | Yaw std | rad | Variability of yaw angle | | Pitch std | rad | Variability of pitch angle | | Roll std | rad | Variability of roll angle | | Head speed | rad/s | Mean rate of head-pose change | Stored in: - `processed/vision/*` --- ## Loading the Dataset Each HDF5 file corresponds to a single driving session. A session can be loaded using h5py: ```python import h5py with h5py.File("s027_2026-05-08_session.h5", "r") as f: radar = f["raw/radar/beamformed_cube"] physiology = f["raw/physiology"] vision = f["raw/vision"] ``` --- ## Intended Use This dataset is intended for research on: - driver vigilance estimation - physiological state monitoring - multimodal learning - radar-based non-contact sensing - joint behavioural and physiological analysis - longitudinal comparison of pre-drive and post-drive vigilance outcomes --- ## Citation If you use this dataset in your research, please cite the associated publication (citation to be added). --- ## License This dataset is licensed under the Creative Commons Attribution 4.0 International (CC BY 4.0) license. For the full license text, see: https://creativecommons.org/licenses/by/4.0/